2021 · United States · O - Public administration and defence · Data exposure via misconfigured low-code web app portals
Analysed 2026-11-07 by Hermes
Open this case in the tool →Data exposure via misconfigured low-code web app portals
Expertise to Execution
Evidence coverage: 3/6 stages, weight 18/60
Decisive stage: entry (clearest evidence) · Alternative: Strategy to Execution
(1) State health agencies ran COVID-19 contact tracing and vaccination portals as public websites under privacy duties that treat survey data as confidential (Indiana runs a privacy policy with a designated privacy coordinator and notified under those duties), and Microsoft's Power Apps portals documentation described exactly when the OData list feeds become publicly accessible and warned to enable table permissions for sensitive information - the technical remedy existed and was documented. (2) In practice the agencies' portals kept the insecure defaults: over a thousand anonymously readable lists across a few hundred portals were found, including 38 million records overall and, for contact tracing alone, roughly 750,000 individuals in Indiana's 'contact' and 'vg_covid19case' lists plus Maryland testing-appointment data; the agencies' own security reviews did not find it, and only UpGuard's external scan (first discovery 24 May 2021) ended the exposure. (3) The classification rests on the exposure/entry stage, which carries the clearest evidence on both sides of the operational test: knowledge sat with technical specialists (a documented configuration remedy, no leadership decision needed) and the chain broke in operations - the configuration was simply never applied, confirmed by the certificate co-signed by the Indiana DOH CIO recording that the state itself misconfigured its portal. That is knowledge within the material column not executed (Expertise to Execution); no source shows leadership deferring, underfunding or accepting the gap, and the secondary Strategy to Execution reading (privacy obligations not operationalised at the data tier) rests on an inference and cannot displace the decisive stage. There was no attacker and no known misuse of the data, so the entry vector here is the misconfiguration itself and is the load-bearing stage; recovery and response were by contrast aligned (immediate correction, records returned, residents notified with credit monitoring). RM27 (Expertise to Execution lens): the exposed services were public health-data portals left on the platform's insecure default settings, and the secure configuration was manual and evidently unmaintainable at the pace these portals were stood up, so the known hardening step was never executed (causal role: Enabling). RM12 (Expertise to Execution lens): the agencies' security reviews passed apps whose data feeds were serving PII anonymously, so the assurance produced no evidence on whether the access controls actually worked and kept the exposure invisible until an external scan (causal role: Obscuring). RM16 (Expertise to Execution lens): the contact tracing portals held crown-jewel personal data of survey respondents with no operational data-tier separation between what should be public and what must stay private, so more and more sensitive data than necessary was exposed to anonymous readers (causal role: Amplifying).
RM27 — Configuration hardening / external exposureRM12 — Assurance / oversight credibility gapRM16 — Data governance / minimisation gapThe Victim column holds a data category, not an organisation: the row's description matches the May-August 2021 Power Apps portals exposure reported by UpGuard; the contact tracing part of it sits with state health portals (Indiana DOH ~749,618 individuals; Maryland Department of Health testing appointments), so Country/Year/Sector are taken from the Indiana incident (US state public health administration). No threat actor and no known misuse of data; the case is an accidental exposure, not an attack. Indiana's press release says UpGuard 'improperly accessed' the data and that the company 'reaches out to seek business'; UpGuard disputes the business solicitation claim, and a certificate co-signed by the Indiana DOH Privacy Officer and CIO records that the state itself misconfigured the portal (cited via UpGuard's report). The victim's own 2021 security posture beyond this portal is not documented; assessment covers the incident lifecycle only. The case file's source_url carries a trailing ' /' fragment; the clean Wired URL was opened instead. Link check: events.in.gov blocks automated access (HTTP 403); the page was opened and read during research. Unattended run: nothing required consent; no leaked data, leak sites or attacker infrastructure was accessed.